How Much Water Should You Drink When Exercising

Should you drink according a plan, or should you drink when you are thirsty? Do we trust our body’s intuition or do we override our instincts with a prescribed drinking schedule? This hydration quandary has sparked a fierce debate within the sports science community.

Here at GU, we believe that our thirst is not always a reliable way to gauge proper hydration during exercise, and we found a recent study that advocates for drinking early and often to achieve optimal performance.

You can find the abstract here, but the cyclists in the study who followed a prescribed hydration plan saw the following benefits:

  • They lost less bodyweight from dehydration
  • They cycled 5% faster during the final hill climb of a 30k time-trial
  • They produced more power in the final 3k of the test
  • They had lower body temperatures after the test

In short, drinking according to a plan resulted in higher cycling speeds, greater power output, and a faster finish time. We like the sound of that!

But why does drinking more improve performance?  

When you lose body fluids during exercise (through sweat, respiration, etc.), your blood plasma volume decreases. Thicker blood is harder for the heart to pump, which leads to two things: reduced blood flow to your working muscles and less heat released through your skin. Dehydration can result in lower cardiac output, less blood flow to your skin, and decreased sweat production, all of which contribute to a rise in core body temperature!

As you might guess, exercising in the heat exacerbates dehydration, which puts even more strain on your body. Hotter temperatures can even shift your body’s fuel preference to burn more carbohydrates, leading to early fatigue if muscle glycogen stores become depleted.

These negative effects can be triggered by as little as 1-2% body weight loss! To make matters worse, exercise also feels harder when you’re dehydrated, as cognition and mood can be negatively affected by even a modest amount of dehydration.

How to hydrate
  • Drink early and often. Start your workout well hydrated by drinking 1/10th your bodyweight in fluid ounces 4 hours prior to starting.
  • During your workout, plan to drink every 15-20 min. If you feel fluids sloshing in your stomach, back off a bit.
  • If you’re a heavy or salty sweater, if the ambient conditions are hot/humid/dry/windy, or if you are training at altitude, then you need to drink more.
  • Practice your drinking strategy and find out what works best for you.
  • While it may not be practical to replace all the fluids you lose during exercise, be sure to rehydrate sufficiently after you’re done by consuming 20-24 oz for every pound of bodyweight lost.


Here’s an easy trick to determine whether or not you are drinking enough by calculating your sweat rate is:

  • Weigh yourself before you exercise
  • Weigh yourself after 1 hour of exercise (towel off any sweat first)
  • Record how much you drank during your workout
  • Calculate your hourly sweat rate with this formula:

16 x [(Starting Weight lbs) – (Weight lbs after 1 hour exercise)] + [fluids consumed during oz] = sweat loss in ounces per hour


“Prescribed Drinking Leads to Better Cycling Performance than Ad Libitum Drinking” (MSSE, June 2017). Carried out in 10 elite, heat-acclimatized cyclists, researchers examined the effects of using a prescribed drinking regimen (PD) to replace calculated sweat losses at 1k intervals, versus drinking to thirst—ad libitum drinking (AD)—during a 30k criterium-like performance test in the heat (31°C/88°F). Cyclists randomly completed the 30k time trial under each condition, PD or AD, which consisted of 3 sets of cycling 5k at 50% maximal power output, followed by a 5k hill climb (3% grade) at all-out effort.


The July edition of MSSE contained scathing commentary from the drink to thirst camp challenging the authors of the study, claiming their statistical interpretation was “erroneous” and “misleading” and should have looked at the 30k trial as a whole, rather than each 5k segment and hill climb. In the same issue, the cycling study authors responded to the commentary, claiming their methods were justified, and that the drink to thirst camp had misread the study and used incorrect calculations to challenge the findings in the first place. Talk about a scientific showdown! Suffice it to say this is likely not the last word on the subject


The American College of Sports Medicine guidelines currently recommends drinking to prevent body weight loss of more than 2%. Yet some scholars contend that prescribed drinking is unnecessary and may even become dangerous, such as during ultra endurance events where hyponatremia (low blood sodium concentration) is one of the most frequent and life-threatening medical issues treated on the course. According to the “drink to thirst” camp, the body is capable of sustaining optimal performance even with exercise-induced bodyweight losses of 5% or more, which has regularly been observed in top Western States 100 finishers.


Bardis, C. N., Kavouras, S. A., Adams, J. D., Geladas, N. D., Panagiotakos, D. B., & Sidossis, L. S. (2017). Prescribed Drinking Leads to Better Cycling Performance than Ad Libitum Drinking. Medicine and Science in Sports and Exercise49(6), 1244-1251.

Goulet, E. D., Gosselin, J., & Hoffman, M. D. (2017). Improper Assessment of the Effect of Ad Libitum Drinking on Cycling Performance. Medicine and Science in Sports and Exercise49(7), 1493.

Hoffman, M. D., Pasternak, A., Rogers, I. R., Khodaee, M., Hill, J. C., Townes, D. A., … & Lipman, G. S. (2014). Medical services at ultra-endurance foot races in remote environments: medical issues and consensus guidelines. Sports Medicine44(8), 1055-1069.

Kavouras, S.A., Bardis, C.N, & Adams, J.D. (2017). Response. Medicine and Science in Sports and Exercise49(7), 1493.

Sawka, M. N., Burke, L. M., Eichner, E. R., Maughan, R. J., Montain, S. J., & Stachenfeld, N. S. (2007). American College of Sports Medicine position stand. Exercise and fluid replacement. Medicine and science in sports and exercise39(2), 377-390.


Back to blog